This function computes the performance index for a real symmetric eigensystem.
EPISF — Performance index. (Output)
NEVAL — Number of eigenvalue/eigenvector pairs on which the performance index computation is based on. (Input)
A — Symmetric matrix of order N. (Input)
EVAL — Vector of length NEVAL containing eigenvalues of A. (Input)
EVEC — N by NEVAL array containing
eigenvectors of A. (Input)
The eigenvector corresponding to the eigenvalue EVAL(J) must be in the
J-th column of
EVEC.
N — Order of the
matrix A.
(Input)
Default: N = size
(A,2).
LDA — Leading
dimension of A
exactly as specified in the dimension statement in the calling
program. (Input)
Default: LDA = size
(A,1).
LDEVEC — Leading
dimension of EVEC exactly as
specified in the dimension statement in the calling program.
(Input)
Default: LDEVEC = size
(EVEC,1).
Generic: EPISF (NEVAL, A, EVAL, EVEC [,…])
Specific: The specific interface names are S_EPISF and D_EPISF.
Single: EPISF (N, NEVAL, A, LDA, EVAL, EVEC, LDEVEC)
Double: The double precision function name is DEPISF.
Let M = NEVAL, l = EVAL, xj = EVEC(*,J), the j-th column of EVEC. Also, let ε be the machine precision, given by AMACH(4), see the Reference chapter of this manual. The performance index, τ, is defined to be
While the exact value of τ is highly machine dependent, the performance of EVCSF is considered excellent if τ < 1, good if 1 ≤ τ ≤ 100, and poor if τ > 100. The performance index was first developed by the EISPACK project at Argonne National Laboratory; see Smith et al. (1976, pages 124− 125).
1. Workspace may be explicitly provided, if desired, by use of E2ISF/DE2ISF. The reference is:
E2ISF (N, NEVAL, A, LDA, EVAL, EVEC, LDEVEC, WORK)
The additional argument is:
WORK — Work array of length N.
E2ISF — Performance Index.
2. Informational errors
Type Code
3 1 Performance index is greater than 100.
3 2 An eigenvector is zero.
3 3 The matrix is zero.
For an example of EPISF, see routine EVCSF.
Visual Numerics, Inc. PHONE: 713.784.3131 FAX:713.781.9260 |